Flora and Vegetation in Different Physiognomies of a Mussununga in Southeastern Brazil

Mussununga is an understudied ecosystem within the Atlantic Forest domain, in sandy spodosol lowlands from Bahia to Espírito Santo. Its physiognomy varies from grassland to forest, with a transitional savannic area. We evaluated the life-form spectra differences between the grassland and savanna Mussunungas and its relationship with the depth of a soil impermeable layer (ortstein). The study area is located in the municipality of Linhares, Espírito Santo State. Ten plots were settled in each physiognomy. The floristic and vegetation spectra (accordingly to Raunkiaer) were compared using the G-test. A total of 35 species into three life-forms were found: Phanerophytes, hemicryptophytes and cryptophytes. The floristic spectra were similar in both physiognomies, with a greater richness of phanerophytes. However, the vegetation spectra of the two areas were different. Phanerophytes dominated in the savannas (where ortstein is deeper), while the shallower ortstein of the grasslands favored hemicryptophytes.


INTRODUCTION
The distribution of plant species is related to environmental variables of ecosystems that limit or favor plant establishment. Little is known about this subject in many tropical environments and the comprehension of this relationship is important to understand the ecosystem functionality (Ribeiro et al., 2009;Paula et al., 2015;König et al., 2019). The Atlantic Forest is the most biodiverse Brazilian biome (BFG, 2018) and some of its ecosystem types are still poorly studied, like the Mussunungas (Saporetti-Júnior et al., 2012).
Mussununga is a kind of ecosystem into the Atlantic Forest domain, occurring in lowlands from southern Bahia to northern Espírito Santo State. The Mussununga physiognomy varies from grasslands to woodlands with sparse, low-statured trees contrasting with the surrounding matrix with dense Ombrophyllous forest (Saporetti-Júnior et al., 2012).
Soils are very important in determining the different Atlantic Forest phytophysiognomies (Magnago et al., 2012;Costa et al., 2016). In the lowlands of the Atlantic Forest of the Southeastern to Northeastern Brazilian coastal plains, the vegetation varies from restingas, tabuleiro forest, swamps and Mussunungas, depending on soil features . Meira-Neto et al. (2005) reported that the Bahia Mussunungas are floristically similar to the restingas, where the dominant soil types, respectively the Spodosols and the Entisols, are both sandy and shallow (Oliveira et al., 2010). However, Meira-Neto et al. (2005) pointed out that the presence of an impermeable layer (ortstein) in the spodosols is the main edaphic feature determining floristic differences between the Mussunungas and the restingas. According to these authors, this impermeable layer restricts root penetration and water percolation, resulting some seasonally flooded sites.
Soil can also print variation in the Mussununga physiognomies, which varies from forest to grassland Saporetti-Júnior et al., 2012;Gastauer et al., 2017). The functionality of the Mussunungas in Bahia was well studied by Saporetti-Júnior et al. (2012). These authors identified different physiognomies with diverse functional group composition determined mainly by the depth of ortstein and soil granulometry features. These authors reported a dominance of therophytes and chamaephytes in grasslands and open savannas, and phanerophytes in the savannas or forest-like Mussunungas. In addition, Meira-Neto et al. (2005) found a higher richness of phanerophytes in Caravelas (Bahia) followed by chamaephytes and hemicryptophytes. Otherwise, in another Bahia Mussununga (Mucuri) the richness of hemicryptophytes was higher due to anthropogenic disturbances, mainly fire and grazing. Heringer et al. (2019) also mentioned changes in the Mussununga vegetation due to anthropogenic impact.
Mussunungas of Espírito Santo State (ES) are not well studied as those in Bahia. Ferreira et al. (2014) reported that they exhibit low floristic similarity. The Mussunungas found in ES also present different phytophysiognomies into two major groups: Mussununga forests and Mussununga grasslands . Araújo et al. (2008) classified the grassland Mussunungas of ES in four types: Graminoid, dense graminoid, closed scrub, and open scrub. According to these authors, this variation seems to be related to the seasonality of flooding events. In ES the grassland Mussununga is known locally as campo nativo or simply nativo (Araújo et al., 2008). Ferreira et al. (2014), also studying ES Mussunungas, reported that the frequency and duration of flooding conditioned floristic differences between grassland and shruby-grassland Mussunungas. These authors found similar floristic spectra in these two phytophysiognomies, with a higher richness of phanerophytes, followed by hemicryptophytes. However, they did not assess the abundance or dominance of the life-forms in order to compare the vegetation spectra. A floristic spectrum is built with a species list, where every species has the same weight. For ecological investigations, assessing the vegetation response is better than merely the flora data. In this case, the species have to be weighted by their abundance and/or dominance to build the vegetation's spectrum (Martins & Batalha, 2011).
Despite the high biological importance of the ES Mussunungas , little is known about its soil-vegetation relationship, highlighting the importance of new studies to understand the functionality of this ecosystem. According to Martins & Batalha (2011), studies of life-forms portray the similarities and peculiarities of environments, which are relevant for the understanding of the ecosystems' structure and function. Ferreira et al. (2014) studied the flora and floristic spectra in ES Mussunungas. However, they did not evaluate the vegetation spectra. The study of Araújo et al. (2008) includes both flora and vegetation, but does not address the functionality aspect (life-forms) associated with the environmental characteristics of the different physiognomies. Despite these previous studies have provided important information about ES Mussunugas, the present study presents, for the first time, aspects of the vegetation (physiognomy), including the species functionality, and their relationship with environment features. As such, this study aimed to assess the effect of the ortstein depth on the flora and vegetation patterns of savannic and grassland Mussunungas of ES, taking into account the occurring species, their life-forms and the respective proportions with which they occur in the communities.
VNR is characterized by the following vegetation formations: High or dense ombrophyllous forests, Mussunungas, marshlands, marsh forests, riparian forests and nativo grasslands that appear as enclaves in the forest areas, occupying approximately 6% of the total reserve area (Araújo et al., 2008). This study was performed at two continuous sampling sites with different Mussununga physiognomies: A savanna and a grassland. The savanna is a transition between the Mussununga forest and the Mussununga grassland, herein denominated savannic Mussununga. In this phytophysiognomy there is an herbaceous layer mixed with shrubs and small trees widely spaced. The trees can reach 10 m in height, but do not form a continuous canopy, similarly to the "open scrub" described by Araújo et al. (2008). The grassland is located in the central portion of the studied area ( Figure 1). This phytophysiognomy is dominated mainly by grasses with a more open shrub layer than the savannas, similarly to the "graminoid type" described by Araújo et al. (2008). In these areas, the seasonal rains (from October to March) bring the water table to the soil surface.
The coastal plain forest in the North of ES is associated with sandy oligotrophic soil of the Barreiras Series (Oliveira et al., 2010). Most of the soils of the Espírito Santo Mussununga grasslands are characterized as Spodosols (Secretti, 2013). According to this Secretti (2013), these soils are sandy, oligotrophic, with a hard and impermeable layer (ortstein) along the B horizon that inhibits water percolation and favors flooding during the rainy season, similarly to those soils described for Bahia Mussunungas (Saporetti-Júnior et al., 2012). The depth at the beginning of the cemented spodic horizon was measured with a ruler in three representative soil profiles in each phytophysiognomy, near to the sampling plots. In the grassland areas, the ortstein depth was lower than 0.6 m (0.36-0.59 m), while in the savannic area the ortstein was deeper than 1 m (1.1-1.5 m).

Sampling design
A transection was established along the two studied physiognomies, at the beginning of the savannic physiognomy (close to the forest border) to the center of the Mussununga patch, with grassland Mussununga. Ten plots of 10 × 10 m each were systematically delimited (10 m equidistant) along this transection for each kind of physiognomy.

Floristic and vegetational studies
All vascular species inside the plots were collected, identified, herborized and deposited at the OUPR and CVRD Herbaria. Species identification was done with specialized literature and by comparison with identified Herbarium specimens.
The species cover was visually estimated as the percent cover of the species on the plot (Messias et al., 2011).
Life-form classification followed the Raunkiaer's original system into the five main classes: Phanerophytes, chamaephytes, hemicryptophytes, cryptophytes and therophytes. In this way, epiphytes and lianas were included into the phanerophyte class. Each species was assigned to only one life-form -the one which the buds were less protected. The floristic spectra represented the richness of species in each life-form. To build the vegetational spectra, the total coverage of individuals in each life-form was considered (Martins & Batalha, 2011).

Data analysis
A species accumulation curve was built using EstimateS Software to estimate the expected total number of species in the study area using Chao-2, Jackknife-1 and Jackknife-2 estimators, as suggested by Moro et al. (2014).
To test if the biological spectra were different in the two Mussununga physiognomies, they were compared pairwise with the Williams G-test.

RESULTS
A total of 35 species into 25 families were sampled. The Bromeliaceae and Eriocaulaceae families presented the greatest richness, with three species each. The Dilleniaceae, Myrtaceae, Orchidaceae, Polypodiaceae, Rubiaceae and Sapindaceae presented two species. The other families presented just one species. Even though the Poaceae family presented only one species, Renvoizea trinii dominated the sampled area with coverage of 80% (Table 1).
The savannic Mussunungas presented a higher species richness (29), being 18 exclusive from this physiognomy. Eighteen species occurred in both physiognomies and six were exclusive from the grasslands.
Three life-forms were found in the grassland and savannic Mussunungas: Phanerophytes, hemicryptophytes and cryptophytes. We did not observe any chamaephytes or therophytes.
The floristic spectra for both phytophysiognomies of the studied Mussunungas were similar (G = 0.7762, p = 0.678) and presented phanerophytes as the predominant life-form, followed by hemicryptophytes ( Figure 2).
On the other hand, there was significant difference (G = 165.13, p < 0.0001) between the vegetational spectra of the two studied phytophysiognomies (Figure 2). In the grassland Mussununga, hemicryptophytes were predominant, followed by phanerophytes and cryptophytes. While in the savannic Mussununga, phanerophytes dominated, followed by hemicryptophytes.

DISCUSSION
Higher species richness was found in the savanna compared with the grassland, where the environmental conditions are more restrictive. Two severe water stresses occur in Mussunungas. First of all, the impermeable cemented soil layer drives waterlogging during the rainy season (Saporetti-Júnior et al., 2012;Ferreira et al., 2014). However, during the dry season, the waterlogging gives place to a severe drought. These environmental constraints in Mussunungas are greater in soils with shallower ortstein (Saporetti-Júnior et al., 2012), limiting the flora and functional diversity. In grassland Mussunungas a thin water sheet could be seen covering the soil surface during the rainy season, when the intensity and periodicity of flooding is greater (Ferreira et al., 2014). This event may be rare or absent in the savannas (Ferreira et al., 2014). According to Meira-Neto et al. (2005), the ortstein depth determines the drainage regime in Mussununga's spodosol. Some species were typical from the grassland Mussunungas, adapted to seasonal flooding like Pagamea guianensis. This species was also described by Araújo et al. (2008) as occurring in areas of graminoid Mussunungas, where the water table comes to the surface. Other woody species like Ocotea notata and Protium icicariba were found only in the savannic Mussununga. Araújo et al. (2008) reported the occurrence of these species along the transition of grassland to Mussununga forests. Other species were found in both physiognomies, showing they were adapted to both conditions, like: Davilla flexuosa, Doliocarpus multiflorus, Humiria balsamifera and Pagamea guianensis, as described by Araújo et al.  (2008). However, these authors reported that there is not a floristic pattern able to distinguish the different physiognomies of no-forest ES Mussunungas.

Family/Species Life-Form Physiogomy C (%) Voucher
In spite of the different Mussununga phytophysiognomies, they presented a similar pattern, with higher richness of phanerophytes followed by hemicryptophytes. Other studies also showed similar floristic spectra both in ES (Ferreira et al., 2014) and Bahia Mussunungas (Meira-Neto et al., 2005). Although Ferreira et al. (2014) have included chamaephytes as a representative life-form in Mussunungas, they used an adapted Raunkiaer System (Martins & Batalha, 2011), in which the height of renewing buds of chamaephytes and phanerophytes is higher than those of the Raunkiaer original method. Thus, many small phanerophytes (by the original Raunkiaer system) would be described as chamaephytes. In addition, by using the original Raunkiaer System, epiphytes and lianas are also considered phanerophytes. Consequently, in this study, four epiphytes (three Bromeliaceae species and a fern) as well as four lianas (Doliocarpus elegans, Smilax rufescens, Urvillea rufescens and Vanilla bahiana) were counted as phanerophytes in the savannic Mussununga. However, most phanerophytes are small trees and shrubs from several families common in the forest-like Mussunungas, like Humiria balsamifera, Guapira pernambucensis (Saporetti-Júnior et al., 2012), Andira nitida and Ocotea notata . Other phanerogamic species are typical from the transitional zone between forest and grassland Mussunungas like Tabebuia elliptica, Protium icicariba and Stigmaphyllum paralias (Araújo et al., 2008).
The lack of nutrients in Mussununga soils may explain the absence of therophytes. Therophytes need a greater nutrient input to trigger their fast growth since they have a short time to complete their cycle (Schmidt, 2012), being almost absent in dystrophic environments (Mioduski & Moro, 2011). A similar pattern was found in the Caatinga (Moro et al., 2016), where areas with eutrophic soils presented a greater therophyte richness. Meira-Neto et al. (2005) and Saporetti-Júnior et al. (2012) reported some therophytes in Bahia Mussunungas. However, the cited Eriocaulaceae and Xyridaceae species are not ephemeral plants and present clonal reproduction by stem buds at the soil level, fitting in the hemicryptophyte description.
The floristic spectrum displays the flora characteristics; that is, the species richness for each life-form, while the vegetational spectrum represents the physiognomy feature. Studies on the distribution of life-forms in tropical environments have evidenced differences in the vegetational spectrum even when there is no difference in floristic spectrum (Batalha & Martins, 2004;Messias et al., 2011).
Saporetti-Júnior et al. (2012) also described a higher dominance of phanerophytes in savannas and forest-like Mussunungas in Bahia. In general, higher proportion of phanerophytes are expected in environments with favorable climates, like in the warm, humid tropical rainforests. On the other hand, harsher climates favor life-forms with greater degree of bud protection (Costa et al., 2016). Analyzing the Brazilian Biomes, Costa et al. (2016) reported that in a macroscale view, climate is the most striking factor determining the biological spectra, however, in a regional scale, soil and anthropogenic disturbance also play an important role in defining the spectra. Even though Mussunungas can be seen as a marginal physiognomy into the Atlantic Forest domain, the pattern of high richness and dominance of phanerophytes is maintained, unless there is another kind of environmental stress limiting the establishment of this life-form. Meira-Neto et al. (2005) reported a Mussununga site (Mucuri, Bahia) with a slight dominance of hemicryptophytes. These authors mentioned that the occurrence of fire and grazing in successive years favored the occurrence of this life-form. Hemicryptophytes are represented mainly by grasses and sedges with a fasciculate root system, exploring the superficial soil layer, and also exhibiting anatomic and physiological adaptations to the seasonal flooding (Visser et al., 2000;Mollard et al., 2008;Hodge et al., 2009;Waring & Maricle, 2012). These abilities explain the high abundance/dominance of hemicryptophytes in grassland Mussunungas, since the upper soil layer is richer in organic matter (Nascimento et al., 2013;Barroso et al., 2014), which increases the nutrient levels and water holding capacity during the dry season. In addition, during the rainy season, when the seasonal flooding occurs, the O 2 concentration in this layer is less limiting (Rich & Watt, 2013). Saporetti-Júnior et al. (2012), studying a Mussununga at Bahia State, found a strong relationship between soil physical characteristics and the abundance of life-forms. The authors reported that the ortstein depth and soil granulometry determine the distribution of functional plant groups. They found that soils with shallower ortstein and coarser granulometry are dominated by therophytes and hemicryptophytes, while phanerophytes dominate in places with deeper ortstein and higher percentage of finer soil particles. Nevertheless, some species described by these authors as therophytes, like Xyris capensis, Actinocephalus ramosus and Syngonanthus nitens, are not annual plants and present clonal reproduction by stem buds at the soil level, fitting in the description of hemicryptophytes. If so, the vegetation spectra are similar in Mussunungas of Bahia and ES, with predominance of hemicryptophytes or phanerophytes, depending on the depth of ortstein.
Thus, the physiognomic differences between the studied Mussununga are influenced by ortstein depth. These results are corroborated by Secretti (2013), who reported that soil and the impermeable layer are deeper near the forested Mussununga, favoring the establishment of taller vegetation in the savannic Mussununga. In addition, in the savannic Mussunungas, in the rainy season the water table rises but does not reach the land surface (Secretti, 2013). According to this author, in the grassland Mussununga, the soil is shallower and the ortstein is more superficial compared to the soils of the forested Mussununga, making it substantially more difficult for taller species to grow there. These factors explain the predominance of the hemicryptophytes, such as the grasses, in the grassland Mussununga, which according Waring & Maricle (2012) and Yamauchi et al. (2018) are adapted to this condition. Although the species richness of the hemicryptophytes is lower than the phanerophytes, Renvoizea trinii (Poaceae) presented a high coverage in these areas, justifying the dominance of hemicryptophytes in the vegetational spectrum. Saporetti-Júnior et al. (2012) observed that a combination of the dry season with flooding episodes is a strong limiting factor that impedes most of the forest species to establish themselves in the Mussunungas. According to these authors, the soil is a limiting factor for the phanerophytes. Araújo et al. (2008) also stated that the stretch of transition between the grassland and forested Mussununga are easily recognized by the substitution of the herbaceous species by trees and can be considered distinct in the phytophysiognomies of the Mussununga. These authors also comment about the low floristic differentiation among the grassland Mussunungas types, and reinforce the idea of the clear difference based on species dominance and vegetation physiognomy. Silva & Batalha (2008) found similar results in floristic spectra in a hyperseasonal cerrado where flooding determined a reduction of species richness and favored the dominance of hemicryptophytes. In campos rupestres, where the soil is sandy, shallow and dystrophic, hemicryptophytes also dominate in seasonal flooding areas (Messias et al., 2013;Silveira et al., 2016). Interestingly, many families common in campos rupestres, such as Eriocaulaceae, Xyridaceae, Poaceae, Orchidaceae, Cyperaceae and Melastomataceae also occur in Mussunungas, which seems to be adapted to these stress conditions. Soil-vegetation relationship is well described for the Mussunungas of Bahia (Saporetti-Júnior et al., 2012). Ferreira et al. (2014) also presented a comprehensive study for Mussunungas of ES. But these authors only contemplated the floristic spectra, not taking into account the dominance or abundance of the functional groups. Since Ferreira et al. (2014) reported that Mussunungas of Bahia and ES exhibit a low floristic similarity, the present study contributes to the comprehension of patterns of biological spectra in Mussunungas. Even though this survey was restricted to a small area, more than one-third of the surveyed species had not been listed before in ES Mussunungas: Anthurium zeneidae, Tillandsia globosa, Actinocephalus ramosus, Neomarica sabinei, Psidium brownianum, Urvillea rufescens, Manilkara triflora, Smilax rufescens and Monteverdia obtusifolia. These results highlight how diverse is the Mussununga flora as well as the importance of focusing on further floristic studies. By analyzing the flora of Mussunungas from different sites (Meira-Neto et al., 2005;Saporetti-Júnior et al., 2012;Ferreira et al., 2014) it seems that Mussunungas have high species turnover (beta diversity), suggesting the need of future research to elucidate that.

CONCLUSION
The environmental constraints resulting from seasonal flooding as well as the physical barriers of the soil influence the vegetational spectra in Mussunungas, conditioning phytophysiognomic differences. Vegetation spectra are similar in Mussunungas of Bahia and Espírito Santo, with a predominance of hemicryptophytes or phanerophytes, depending on the ortstein depth. Phanerophytes dominated in the savannas (where ortstein is deeper), while the shallower ortstein of the grasslands favored hemicryptophytes.
The results presented contribute to the knowledge of the flora, physiognomy and functional diversity patterns in Mussunungas, as well as in other ecosystems with seasonal flooding.